all-MiniLM-L6-v2-quora

This model is a fine-tuned version of sentence-transformers/all-MiniLM-L6-v2 on the quora dataset. It achieves the following results on the evaluation set:

• Loss: 0.0865
• Accuracy: 0.8150
• F1: 0.7945

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 5e-05
• train_batch_size: 32
• eval_batch_size: 64
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1
0.087	1.0	11371	0.0829	0.4143	0.5535
0.0794	2.0	22742	0.0783	0.6017	0.6458
0.0606	3.0	34113	0.0756	0.3631	0.5327
0.05	4.0	45484	0.0781	0.4475	0.5679
0.0448	5.0	56855	0.0789	0.6856	0.6975
0.0372	6.0	68226	0.0761	0.3922	0.5443
0.033	7.0	79597	0.0786	0.7586	0.7494
0.032	8.0	90968	0.0780	0.5011	0.5927
0.0229	9.0	102339	0.0819	0.7513	0.7439
0.0198	10.0	113710	0.0840	0.5522	0.6185
0.0169	11.0	125081	0.0821	0.7959	0.7785
0.0199	12.0	136452	0.0807	0.8353	0.8118
0.0118	13.0	147823	0.0819	0.8418	0.8176
0.0123	14.0	159194	0.0816	0.7577	0.7487
0.0093	15.0	170565	0.0856	0.7934	0.7765
0.0124	16.0	181936	0.0843	0.8484	0.8241
0.008	17.0	193307	0.0838	0.7998	0.7818
0.0106	18.0	204678	0.0872	0.8245	0.8027
0.0066	19.0	216049	0.0857	0.8122	0.7922
0.0059	20.0	227420	0.0865	0.8150	0.7945

Framework versions

• Transformers 4.21.3
• Pytorch 1.12.1+cu116
• Datasets 2.5.1
• Tokenizers 0.12.1